Nano language and distribution of article title terms according to power laws
Explicit semantic context directly furthers understanding of content in humans. Applications range from efficient information exchange in meetings to visual information retrieval or orientation in document collections. The automatically extended context addresses various issues simultaneously: a problem is depicted from different points of view, new ideas are associated to those already mentioned, the trail lends an assisting and moderating functionality to a discussion and finally the network of concepts can be saved or printed out for later reference. Obviously, discussions only benefit from this context, if provided in real-time. In this contribution we present the software “SemanticTalk”, which provides access to the underlying conceptual structures of unstructured speech streams. Essentially, the most significant concepts of a speech stream are extracted and visualized and related to further associated concepts automatically retrieved from a collection of documents. This results in an incrementally growing conceptual graph, which is dynamically unfolded in real-time.